Skip to main content

An OCR library for Python

Project description

LanyOCR

A general OCR to detect and recognize English texts in an image based on a combination of EasyOCR and PaddleOCR.

LanyOCR automatically merges text boxes into lines even for rotated texts.

alt text

alt text

Getting Started

Install dependencies

pip install lanyocr

Run example

PYTHONPATH=. python detect.py --merge_rotated_boxes true --merge_vertical true --image_path images/example1.jpg

Faster version, a bit less accurate

PYTHONPATH=. python detect.py --merge_rotated_boxes true --merge_vertical true --merge_boxes_inference true --image_path images/example1.jpg

Switch to different recognizer

PYTHONPATH=. python detect.py --merge_rotated_boxes true --merge_vertical true --recognizer_name paddleocr_en_mobile --image_path images/example1.jpg

Recognize other languages

PYTHONPATH=. python detect.py --merge_rotated_boxes true --merge_vertical true --recognizer_name paddleocr_french_mobile --image_path images/french_example1.jpg

Output image will be in outputs/output.jpg

Supported Languages

  • English: paddleocr_en_server, paddleocr_en_mobile
  • French: paddleocr_french_mobile
  • Latin: paddleocr_latin_mobile

Note: Some unicode characters cannot be visualized correctly by OpenCV, please find the text lines in the console log.

Validate accuracy

Download ICDAR 2015 dataset

bash datasets/download_icdar2015.sh

Validate accuracy

python benchmark.py

Online API

You can try LanyOCR free on RapidAPI

To Do

[x] Abstract Class/Interface for each component
    [x] LanyOcrDetector: outputs locations of text boxes        
    [x] LanyOcrMerger: merge text boxes into text lines
    [x] LanyOcrRecognizer: convert text boxes/lines into text
    [x] LanyOcrAngleClassifier: estimate the angle of a text box/line

[ ] Multi-languages support
    [X] French        
    [X] Latin
    [ ] German

[ ] Inference using multi-models to improve accuracy
    [ ] Add interface to support voting policy

[ ] Expose flags to configure each component in OCR pipeline

Known issues

[ ] Visualization step: some small texts are drawn in incorrect directions

License

This project is licensed under the MIT License.

Credits

Special thanks to authors and developers of EasyOCR and PaddleOCR projects.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

lanyocr-0.1.3-py3-none-any.whl (86.4 kB view details)

Uploaded Python 3

File details

Details for the file lanyocr-0.1.3-py3-none-any.whl.

File metadata

  • Download URL: lanyocr-0.1.3-py3-none-any.whl
  • Upload date:
  • Size: 86.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/4.0.2 CPython/3.9.12

File hashes

Hashes for lanyocr-0.1.3-py3-none-any.whl
Algorithm Hash digest
SHA256 7219b8f81e67c06f272be187ed23d4de88953df03a9cf178b9d78d76bc130a27
MD5 4aa6718750741e082a50d0ceaaf5e4eb
BLAKE2b-256 a1fd0e98fc95836c486f165b5b9fae5fbded36a4614f798ff2da0b0cc7dc7ce0

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page